TW201301874A - Method and device of document scanning and portable electronic device - Google Patents

Abstract

A document scanning method includes receiving a plurality of images of a plurality of blocks of a document by an image receiver, adjusting characteristics of the plurality of images according to distances between the image receiver and the plurality of blocks when the image receiver is receiving the plurality of images, to generate a plurality of adjusted images, determining a plurality of characteristic points of each of the plurality of adjusted images and a plurality of characteristic vectors corresponding to the plurality of characteristic points, and combining the plurality of adjusted images according to the plurality of characteristic vectors corresponding to the plurality of characteristic points of each of the plurality of adjusted images, to generate a scanning result.

Description

Document scanning method, document scanning device and portable electronic device

The invention relates to a document scanning method, a document scanning device and a portable electronic device, in particular to a document scanning method, a document scanning device and a portable electronic device capable of combining different images according to image feature points to provide a document scanning function. Device.

Portable electronic devices, such as notebook computers, lithographic computers, smart phones, etc., have the characteristics of small size, light weight, and easy portability, so that users can have powerful computing power and paper processing functions anytime, anywhere. Has become one of the necessary tools for business people. Under this circumstance, how to improve the functions of portable electronic devices to meet various needs has become one of the goals of the industry.

For example, business people usually carry a laptop when they go out to display product information or record meeting materials. However, when it is temporarily encountered that a paper document needs to be scanned into an electronic file, since the scanner is not carried with the notebook computer in general, the user often needs to go to a nearby convenience store or return to the company. Scanning again, not only costs money and time, but also misses the immediacy of this information. Therefore, it is much more convenient if the notebook has the function of scanning files.

Therefore, the present invention mainly provides a document scanning method, a document scanning device, and a portable electronic device.

The present invention discloses a file scanning method, which includes a plurality of images of a plurality of blocks of a file captured by an image capturing device; and the image capturing device captures the plurality of images and the plurality of blocks according to the image capturing device a distance, adjusting a characteristic of the plurality of images to generate a plurality of adjusted images; determining a plurality of feature points of each adjusted image in the plurality of adjusted images and a plurality of feature vectors of the plurality of feature points; The plurality of feature vectors of each adjusted image of the plurality of adjusted images are combined with the plurality of adjusted images to generate a scan result of the file.

The invention further discloses a document scanning device, comprising: an image capturing device; a ranging unit; a processor; and a storage unit, wherein the storage unit is configured to store a code, the code indicating that the processor performs the following steps Controlling the image capturing device to capture a plurality of images of a plurality of blocks of a file; controlling the ranging unit to measure a distance from the plurality of blocks when the image capturing device captures the plurality of images; The image capture device captures the distance from the plurality of blocks when the plurality of images are captured, and adjusts the characteristics of the plurality of images to generate a plurality of adjusted images; and determines each of the adjusted images in the plurality of adjusted images. a plurality of feature points and a plurality of feature vectors of the plurality of feature points; and combining the plurality of adjusted images according to the plurality of feature vectors of each adjusted image of the plurality of adjusted images to generate the file A scan result.

The present invention further discloses a portable electronic device including a processor, a storage unit, an image capture device, and a file scanning device, including a ranging unit, and a code stored in the storage unit. The code indicates that the processor performs the following steps: controlling the image capturing device to capture a plurality of images of a plurality of blocks of a file; and controlling the ranging unit to measure the image capturing device to capture the plurality of images The distance between the plurality of blocks; the distance between the plurality of blocks when the image capturing device captures the plurality of blocks, and adjusting the characteristics of the plurality of images to generate a plurality of adjusted images; determining the plurality of adjustments a plurality of feature points of each adjusted image in the post-image and a plurality of feature vectors of the plurality of feature points; and the plurality of feature vectors according to each adjusted image of the plurality of adjusted images, combined with the plurality of feature vectors The adjusted image is used to produce a scan of the file.

Portable electronic devices such as notebook computers, lithographic computers, and smart phones are usually equipped with a camera for providing image capture functions such as photography, photography, and video telephony. In this case, the present invention utilizes the image capturing function of the portable electronic device to implement the scanning function, and cooperates with a series of calculation processes to effectively combine different image portions of the same file into a complete file or image data.

Please refer to FIG. 1. FIG. 1 is a schematic diagram of a document scanning device 10 according to an embodiment of the present invention. The file scanning device 10 can be disposed in a portable electronic device such as a notebook computer, a lithographic computer, or a smart phone, and is provided by an image capturing device 100, a ranging unit 102, a prompting unit 114, and a processing module 104. Composed of. The image capturing device 100 can be a camera, a video device, or the like originally set by the portable electronic device for capturing images. The ranging unit 102 measures the actual distance from the image capturing device 100 when capturing an image by using a distance measuring principle such as infrared rays or ultrasonic waves. The prompting unit 114 can display the ranging result of the ranging unit 102, which can be a light number, a sound, or a message displayed on a screen. The processing module 104 is composed of a processor 106 and a storage unit 108. The storage unit 108 stores a file scanning code 110 for instructing the processor 106 to perform a file scanning function.

When the file scanning function is to be performed, the image scanning device 10 captures the images IMG_1 to IMG_n of the blocks BLK_1 to BLK_n of the file to be scanned 112 by the image capturing device 100, and captures the image IMG_1~ at the image capturing device 100. In the case of IMG_n, the ranging unit 102 measures the distances DT_1 DT DT_n of the image capturing device 100 and the blocks BLK_1 ～ BLK_n, and displays the corresponding measurement results through the prompting unit 114. The processing module 104 first adjusts the characteristics of the images IMG_1 to IMG_n according to the distances DT_1 ～ DT_n, and then determines the feature points and feature vectors therein, and finally combines the adjusted images to generate the scan result SCN of the file 112.

Regarding the operation principle of the processing module 104, the following steps are described in detail.

First, image adjustment:

As described above, the image capturing device 100 may be a camera or the like originally set by the portable electronic device, or may be an additional device. The space of the portable electronic device is small, and the image capturing quality is maintained. Therefore, the file scanning device 10 sequentially extracts the file 112 according to different blocks BLK_1 to BLK_n. It should be noted that the adjacent blocks in the blocks BLK_1 to BLK_n need to be partially overlapped, and image combination can be performed at the beginning, and the processing manners when how to combine images and when they cannot be combined will be described in detail later. In addition, when the images IMG_1~IMG_n of the blocks BLK_1-BLK_n are captured, it is preferably manually translated by the user, but is not limited thereto, and the prompting unit 114 can be the most to ensure that the user's translation mode conforms to the system requirements. The distance is informed to the user to avoid situations of varying sizes.

Therefore, when the file 112 is scanned, the user moves the file 112 to be scanned before the image capturing device 100, and the image capturing device 100 continuously captures the image IMG_1 of the blocks BLK_1 to BLK_n of the file 112. IMG_n. Since the handheld mode is carried by the file 112, the distances of the images IMG_1 to IMG_n that are captured due to external environmental factors are inconsistent, and the sizes of the images IMG_1 to IMG_n may be different, which may result in the files after the bonding. There will be significant block differences or step differences in the image content. Therefore, the vertical distance between the file 112 and the image capturing device 100 is quite important, and too far or too close will affect the sharpness of the captured text content. In this case, when the user performs scanning, the ranging unit 102 can notify the user of the value of the optimal shooting distance between each block of the file 112 and the image capturing device 100 through the prompting unit 114.

In addition, while the image capturing device 100 captures the images IMG_1 to IMG_n, the ranging unit 102 also synchronously records the vertical distances DT_1 to DT_n between the blocks BLK_1 to BLK_n and the image capturing device 100 at each shooting. The recorded distances DT_1 to DT_n are the basis for performing an important reference for post-processing fine-tuning such as image reduction, enlargement, sharpening, and contrast adjustment when the images IMG_1 to IMG_n of the subsequent blocks BLK_1 to BLK_n are next. At the same time, if the processing module 104 detects that the image of a certain block of the file 112 is not good or can not be used, the user may be recommended to perform the re-shooting action only for the block, so that the final stitching is performed. The content of the image will be more consistent, and the quality of the finished product will be better.

Therefore, when acquiring the images IMG_1 - IMG_n, the processing module 104 records the distances DT_1 - DT_n between the corresponding blocks BLK_1 - BLK_n and the image capturing device 100 as the basis for scaling and adjusting the images IMG_1 - IMG_n. . There are many ways to adjust the images IMG_1 to IMG_n. For example, the average value of the distance (DT_2, DT_4, ...) at the even-numbered blocks (BLK_2, BLK_4...) can be used as the reference point. , as a scaling benchmark. For example, if the images IMG_1 to IMG_4 of the four blocks BLK_1 to BLK_4 are taken in total, and the shooting distance DT_2 of the second image IMG_2 is 2 cm, the shooting distance DT_4 of the fourth image IMG_4 is 3 cm, and the average of the two is 2.5 cm. As a zoom reference; therefore, the second image IMG_2 needs to be enlarged by 2/2.5=0.8 times, while the fourth image IMG_4 needs to be enlarged by 3/2.5=1.2 times.

In addition, further, bilinear interpolation can be used as an algorithm for scaling processing. After all the captured images are zoomed and adjusted to the same reference, the image joint processing of each block of the file is performed.

Second, the image merger:

The basic concept of image merging is to first find the same image content in adjacent images, thereby combining the images. In order to judge the content of the image, a representative feature point needs to be selected first, and usually the image has many types of features, such as texture, color, shape, contour, etc., and the present invention searches for representative corner points in the image ( Corner point) feature.

(a) Looking for feature points:

In order to judge the corner feature, the integral image technique is first used to speed up the calculation, and the quantized Gaussian function second-order partial differential is used to calculate the Hessian matrix determinant value of the image. The variation σ produces different Hays determinant matrices to find feature points.

In detail, if I ( x , y ) represents a pixel of coordinates ( x , y ) in an image I , the Hayes matrix of this point is represented by the variability of σ. Where h ( x , y , σ) is the Hayesian matrix corresponding to the pixel I ( x , y ); L _xx ( x , y , σ) is the pixel I ( x , y ) and the Gaussian function g (σ) Second derivative in the x-axis direction The gyro product; L _xy ( x , y , σ ) and L _yy ( x , y , σ) are the pixels I ( x , y ) and The value of the convolution product is performed, and g (σ) is a Gaussian function. Each element H ( x , y , σ) is the determinant value of the Hayes matrix of the image I at coordinates ( x , y ), which can be written as H ( x , y , σ ) = L _xx ( x , y , σ * L _yy ( x , y , σ) - ( L _xy ( x , y , σ)) ² , which can be called H is a Hayesian matrix of the image I corresponding to the σ value. Different Hays determinant matrices H can be obtained by using different σ values. For example, if σ ₁ =1.2, σ ₂ =2, and σ ₃ =2.8, H ₁ , H ₂ , and H _{3 in} different scale spaces can be obtained. When the number of images I corresponds to different σ values of the Hayes determinant matrix, then the feature points are searched from these Hays determinant matrices.

Suppose there is a pixel whose coordinates are ( x , y ), which is the X point in Fig. 2, and H ₁ , H ₂ , and H ₃ are the Hayes determinant matrix corresponding to σ1, σ2, and σ3 in Fig. ₂ , respectively. Then X points its 26 adjacent pixels (the nine adjacent points of the upper and lower scales and the eight adjacent points of its own scale) are as shown by the dots in Figure 2. If the X point has the largest Hays determinant value among its 26 neighboring pixels, the X point is called the feature point.

(b) Determine the feature vector:

After finding the feature points, since a single feature point cannot accurately describe the data of the area around the feature point, it is necessary to draw a neighboring area near the feature point, and use the adjacent area to generate a feature vector representing the feature point. First, before calculating the feature vector describing a feature point X, in order to make the feature vector resistant to rotation, that is, after the image is rotated, the feature points of the two images can be effectively found, so it is necessary to find the feature point X. The main direction. Therefore, it is possible to draw a region from the neighboring square region (such as 20*20) with the feature point X as the center as R, and to circulate the region R with the Haar horizontal and vertical filters, respectively, and dx. ( x , y ) and dy ( x , y ) represent the result of completing the convolution product, and dx ( x , y ), dy ( x , y ) are matrices of the same size as the region R (eg 20*20). Therefore, for each pixel in the region R, there are corresponding values in dx ( x , y ) and dy ( x , y ).

Then, the feature point X can be centered, and 0 to 60 degrees is an area. The total of all the pixels dx ( x , y ) and dy ( x , y ) in the region can be calculated to obtain a sequence pair. ), that is, (Σ dx ( x , y ), dy dy ( x , y )), using this sequence as a vector, calculating the length of the vector L = This is the length of the region from 0 to 60 degrees centered on the feature point X. Next, lengths of 60 degrees to 120 degrees, 120 degrees to 180, 180 degrees to 240 degrees, 240 to 300 degrees, and 300 degrees to 360 degrees, respectively, are calculated. Assuming that it has the highest length between 60 and 120 after statistics, the main direction of the feature point X can be set to 90 degrees (because (120+60)/2). In this example, all pixels in the region R are counted every 60 degrees, but in practical applications, the pixels in the region R can be counted in a smaller interval to find a more correct direction. In other words, the choice of angle range will affect the ability to accurately find the main direction of the feature point.

After the main direction of the feature point X is obtained, the main direction of the feature point X is dominant, and the main direction of the image is rotated to the north, and a new region R' is defined, wherein the region R' is also the feature point X. The centered area (such as the square area of 20*20) is designed to generate a eigenvector with rotation invariance, and the result is represented by dx '( x , y ), dy '( x , y ), respectively. Dx '( x , y ), dy '( x , y ) is also a matrix of the same size as the region R'.

Next, the region R' is divided into a plurality of sub-blocks, such as 4*4, each sub-block size is 5*5, and each of the sub-blocks is represented by dx '( x , y ) and dy '( x , y ) The block calculates the following four components, (Σ dx '( x , y ), Σ dy '( x , y ), Σ| dx '( x , y )|, Σ∣ dy '( x , y )|), And four components represent one sub-block. In this example, since the region R' has 4*4 blocks, each sub-block has four components, so it can be represented by the main direction of the feature point X, from left to right and top to bottom respectively. The four components of each sub-area are concatenated to form a 64-dimensional feature vector, and this feature vector is used to represent the feature point X.

(c) Image comparison and merger:

When comparing two images, firstly, all the feature points found from the two images are generated by the feature vectors of each feature point through the above (b) method, and the feature vectors of the feature points are compared with each other. And find the most consistent feature points. The comparison method is as follows. If X is a feature point, the Euclidean distance between the feature vector of the feature point X and the feature vector of the other feature points is calculated during the alignment of the feature points. It is assumed that the feature points X1 and X2 are the feature points closest to X and closest to X, and the distances are d 1 and d 2 respectively. If d 1 < r * d 2 , it is determined that X1 and X are the most consistent, where r is self. Set the coefficient, such as 0.5. Finally, the two closest feature points are superimposed, that is, the merging of the two images is completed, which is why it is necessary to ensure that the two adjacent images partially overlap.

(d) Color conversion:

Before starting to judge the feature points, the present invention can also perform pre-processing on each block image, so that the similar color in the image is quantized into the same color, so as to extract the global features more effectively, the main purpose is to let Image content with similar similarity can be classified into the same category to reduce the situation where the two images have similar content when the image is merged, but the content of the two is judged to be different due to a slight difference in light. There are many ways to convert colors. For example, you can convert images from RGB color space to CIEL*a*b* color space, and then quantize the two sub-frequency of a* and b* in image CIEL*a*b*. Finally, convert the image from CIEL*a*b* back to the RGB color space.

The purpose of color quantization is to reduce the difference in image color after quantization, that is, the colors with little difference are normalized to the same color. In order to allow users to have the correct accuracy and efficiency when scanning documents, the selection of features must be based on the stability of feature extraction or the processing speed. Therefore, the principle of selecting features is between block images. When the difference is large, the difference between the features can be clearly indicated. Conversely, when the block image is only slightly different, it is ignored or made inconspicuous. In this case, simply using the corner feature of (a) above does not provide a more accurate discriminating power, so it is necessary to add more features describing the details of the image to combine with the above features to represent the feature vector of the image, for example, available. Sobel edge features, invariant moments, standard deviations and mean features of RGB colors, etc., combined with the aforementioned corner features into one feature vector, and each feature point of each image is generated. The feature vector lengths are the same, and are used as the basis for merging images. In other words, the present invention can use the method of the above (a) to find the feature point, and select a square area (such as 15*15) with the feature point as the center circle, and calculate the Sobel characteristics of the area for the square area respectively. (225 dimensions), 7 invariant moment features (7 dimensions), and RGB color features (6 dimensions), and finally obtain a feature vector with a dimension of 238 dimensions.

After the color conversion process is added, the operation mode of the related image comparison and combination is still in accordance with the method of the foregoing (c), and will not be described herein.

(e) Debugging:

If in the process of processing the merge, there is any case where the feature points of the image cannot be merged with the feature points of other images, or if the matching feature points are less than a set value (such as 20), Then, the document scanning device 10 of the present invention can issue a warning through the prompting unit 114, and let the user re-shoot the corresponding block on the file 112, and then find all the feature points and feature vectors of the retake image, and other The previously merged parts are merged.

Therefore, as apparent from the above description, the document scanning apparatus 10 of the present invention can divide the file 112 into the blocks BLK_1 to BLK_n for photographing, and finally merge them into the scan result SCN through a series of algorithms. At the same time, when the merge is performed, if the merge fails, the debug can be effectively performed to ensure the correctness of the scan result SCN.

The above described operation of the document scanning device 10 can be further summarized into a document scanning process 30, as shown in FIG. The file scanning process 30 includes the following steps:

Step 300: Start.

Step 302: The image capturing device 100 captures the images IMG_1 - IMG_n of the blocks BLK_1 - BLK_n of the file 112.

Step 304: The distance measuring unit 102 measures distances DT_1 DT DT_n from the blocks BLK_1 ～ BLK_n when the image capturing device 100 captures the images IMG_1 ～ IMG _n.

Step 306: The processing module 104 adjusts the characteristics of the images IMG_1 - IMG_n according to the distances DT_1 - DT_n.

Step 308: The processing module 104 determines the feature points of each of the adjusted images IMG_1 to IMG_n and the corresponding feature vectors.

Step 310: The processing module 104 combines the adjusted images IMG_1 - IMG_n according to the feature vectors of the adjusted images IMG_1 - IMG_n to generate the scan result SCN of the file 112.

Step 312: End.

The file scanning process 30 is a summary of the operation mode of the file scanning device 10, detailed operations and changes, such as methods for determining feature points and feature vectors, debugging mechanisms, etc., refer to the foregoing. In addition, in the file scanning device 10, all operational logic is compiled into the file scanning program code 110, which should be the intrinsic knowledge of the computer software for those skilled in the art. However, it should be noted that the file scanning program code 110 is not limited to any programming language, and any programming language that can be executed by the processor 106 to achieve a corresponding function can be used in the present invention.

Furthermore, each component of the document scanning device 10 is the most basic component of the present invention. However, depending on the requirements of different systems, each component may be the original equipment of the system, and the invention is implemented by being transferred. For example, a portable electronic device such as a notebook computer or a smart phone is originally equipped with a central processing unit, a memory, a camera, etc., which can be used to implement the processor 106, the storage unit 108, and the image capturing device of the document scanning device 10. 100. Similarly, the ranging unit 102 can also be implemented by a Proximity Sensor, and the prompting unit 114 can be implemented by a message window on the screen, a sound emitted by a speaker, or the like. In other words, when the file scanning device 10 is applied to the portable electronic device, the file scanning program code 110 may only be added to implement the corresponding function, so that the user's willingness to use is effectively improved.

In the prior art, when a temporary need to scan a paper document into an electronic file, since the scanner is not carried with the notebook computer in general, the user often needs to go to a nearby convenience store. Or after scanning back to the company, it will not only cost money and time, but also miss the immediacy of this information. In this case, the present invention can provide a file scanning function in the portable electronic device, thereby effectively improving convenience.

In summary, the present invention combines different images according to image feature points, thereby providing a file scanning function in a portable electronic device to improve the functionality and convenience of the portable electronic device.

The above are only the preferred embodiments of the present invention, and all changes and modifications made to the scope of the present invention should be within the scope of the present invention.

10. . . Document scanning device

100. . . Image capture device

102. . . Ranging unit

104. . . Processing module

106. . . processor

108. . . Storage unit

110. . . File scanning code

112. . . file

114. . . Prompt unit

BLK_1～BLK_n. . . Block

IMG_1~IMG_n. . . image

DT_1~DT_n. . . distance

SCN. . . Scan result

X. . . Feature points

30. . . File scanning process

300, 302, 304, 306, 308, 310, 312. . . step

FIG. 1 is a schematic diagram of a document scanning device according to an embodiment of the present invention.

2 is a schematic diagram of a Hays determinant matrix according to an embodiment of the present invention.

FIG. 3 is a schematic diagram of a file scanning process according to an embodiment of the present invention.

10. . . Document scanning device

100. . . Image capture device

102. . . Ranging unit

104. . . Processing module

106. . . processor

108. . . Storage unit

110. . . File scanning code

112. . . file

114. . . Prompt unit

BLK_1～BLK_n. . . Block

IMG_1~IMG_n. . . image

DT_1~DT_n. . . distance

SCN. . . Scan result

Claims (17)

A file scanning method includes: capturing, by an image capturing device, a plurality of images of a plurality of blocks of a file; and extracting, by the image capturing device, a distance from the plurality of blocks when capturing the plurality of images, Adjusting characteristics of the plurality of images to generate a plurality of adjusted images; determining a plurality of feature points of each adjusted image in the plurality of adjusted images and a plurality of feature vectors of the plurality of feature points; and according to the complex number The plurality of feature vectors of each adjusted image of the adjusted image are combined with the plurality of adjusted images to generate a scan result of the file. The document scanning method of claim 1, wherein the characteristics of the plurality of images are selected from the group consisting of magnification, sharpness, contrast, and chroma. The file scanning method of claim 1, wherein the step of determining a plurality of feature points of each of the plurality of adjusted images and a plurality of feature vectors of the plurality of feature points includes: calculating each Each pixel of the adjusted image corresponds to a plurality of Hessian matrices of a plurality of variograms; and the plurality of features are selected from the plurality of pixels according to a Hayes matrix of a plurality of pixels of each adjusted image a Haar operation result of calculating a plurality of neighboring pixels of each of the plurality of feature points; and a result of calculating a plurality of neighboring pixels according to each feature point of the plurality of feature points And determining the plurality of feature vectors. The file scanning method of claim 1, wherein the step of generating the scan result of the file is performed according to the plurality of feature vectors of each adjusted image of the plurality of adjusted images combined with the plurality of adjusted images The method includes: comparing the plurality of feature vectors of each adjusted image of the plurality of adjusted images to determine similar feature points in the plurality of adjusted images; and similar features in the plurality of adjusted images The point is a reference, and the plurality of adjusted images are sequentially combined to generate the scan result of the file. The method for scanning a document according to claim 4, further comprising: retrieving one of the adjusted images after the adjusted image has no similar feature points between the adjusted image and the other adjusted images; The image of the block. A file scanning device includes: an image capturing device; a ranging unit; a processor; and a storage unit for storing a code, the code indicating that the processor performs the following steps: controlling The image capturing device captures a plurality of images of a plurality of blocks of a file; and the distance measuring unit controls the distance of the plurality of blocks when the image capturing device captures the plurality of images; according to the image Taking the distance between the plurality of blocks and the plurality of blocks, adjusting the characteristics of the plurality of images to generate a plurality of adjusted images; determining a plurality of the adjusted images in the plurality of adjusted images a feature point and a plurality of feature vectors of the plurality of feature points; and combining the plurality of adjusted images according to the plurality of feature vectors of each adjusted image of the plurality of adjusted images to generate a scan of the file result. The document scanning device of claim 6, wherein the plurality of images are selected from the group consisting of magnification, sharpness, contrast, and chroma. The file scanning device of claim 6, wherein the step of determining a plurality of feature points of each of the plurality of adjusted images and a plurality of feature vectors of the plurality of feature points comprises: calculating each Each pixel of the adjusted image corresponds to a plurality of Hessian matrices of a plurality of variograms; and the plurality of features are selected from the plurality of pixels according to a Hayes matrix of a plurality of pixels of each adjusted image a Haar operation result of calculating a plurality of neighboring pixels of each of the plurality of feature points; and a result of calculating a plurality of neighboring pixels according to each feature point of the plurality of feature points And determining the plurality of feature vectors. The document scanning device of claim 6, wherein the step of generating the scan result of the file is performed according to the plurality of feature vectors of each adjusted image of the plurality of adjusted images combined with the plurality of adjusted images The method includes: comparing the plurality of feature vectors of each adjusted image of the plurality of adjusted images to determine similar feature points in the plurality of adjusted images; and similar features in the plurality of adjusted images The point is a reference, and the plurality of adjusted images are sequentially combined to generate the scan result of the file. The document scanning device of claim 9, further comprising: in the plurality of adjusted images, if one of the adjusted images has no similar feature points between the other adjusted images, retrieving one of the adjusted images The image of the block. The file scanning device of claim 6, further comprising a prompting unit, the code further instructing the processor to instruct the user to translate the file through the prompting unit, so that the image capturing device sequentially captures the plurality The plurality of images of the blocks. A portable electronic device includes: a processor; a storage unit; an image capture device; and a file scanning device, including: a ranging unit; a code stored in the storage unit, The code indicates that the processor performs the following steps: controlling the image capturing device to capture a plurality of images of a plurality of blocks of a file; and controlling the ranging unit to measure the image capturing device to capture the plurality of images The distance between the plurality of blocks; the distance between the plurality of blocks when the image capturing device captures the plurality of blocks, and adjusting the characteristics of the plurality of images to generate a plurality of adjusted images; determining the plurality of adjustments a plurality of feature points of each adjusted image in the post-image and a plurality of feature vectors of the plurality of feature points; and the plurality of feature vectors according to each adjusted image of the plurality of adjusted images, combined with the plurality of feature vectors The adjusted image is used to produce a scan of the file. The portable electronic device of claim 12, wherein the plurality of images are selected from the group consisting of magnification, sharpness, contrast, and chroma. The portable electronic device of claim 12, wherein the step of determining a plurality of feature points of each of the plurality of adjusted images and a plurality of feature vectors of the plurality of feature points comprises: calculating Each pixel of each adjusted image corresponds to a plurality of Hessian matrices of a plurality of variograms; and the complex number is selected from the plurality of pixels according to a Hayes matrix of a plurality of pixels of each adjusted image a feature point; a Haar operation result of a plurality of neighboring pixels of each of the plurality of feature points; and a plurality of neighboring pixels of each feature point of the plurality of feature points Calculate the result and determine the plurality of feature vectors. The portable electronic device of claim 12, wherein the plurality of eigenvectors of each adjusted image of the plurality of adjusted images are combined with the plurality of adjusted images to generate the scan result of the file. The step of: comparing the plurality of feature vectors of each adjusted image of the plurality of adjusted images to determine similar feature points in the plurality of adjusted images; and comparing the plurality of adjusted images The feature point is a reference, and the plurality of adjusted images are sequentially combined to generate the scan result of the file. The portable electronic device of claim 15 further comprising: when the adjusted image has no similar feature points between the adjusted image and the other adjusted images, retrieving the adjusted image An image of one of the blocks. The portable electronic device of claim 12, further comprising a prompting unit, the code further instructing the processor to instruct the user to translate the file through the prompting unit, so that the image capturing device sequentially captures The plurality of images of the plurality of blocks.